The Limitations of Behavioral Biopolitical Science (2)

In my previous post, I identified Man Is by Nature a Political Animal, edited by Peter Hatemi and Rose McDermott, as a good survey of the research in behavioral biopolitical science.  This research constitutes a biological study of politics guided by the intellectual commitments of behavioralist political science, which assumes that any true science of politics must be a value-free science based on abstract causal models of behavior and empirical hypothesis-testing.  Much of this research is fascinating in showing how human biology shapes human politics in ways that have often been discussed on this blog.

But there are also some limitations to behavioralist biopolitics.  While it might explain some of the psychological preconditions for political experience, it does not explain the specific content of political experience as a moral debate over the common good.  Because of its traditional commitment to the fact-value dichotomy and the idea that science must be value-free, behavioralist biopolitical science is not empirical enough in its research, because it must ignore the empirical data of political speech and argumentation about the moral issues of political life.  In doing this, it must ignore the normative structure of animal movement as purposeful conduct.

By contrast, I have indicated my agreement with Ed Wilson that political science is "primarily the study of applied ethics," and thus a biologically grounded science of politics must study the biology of the moral sentiments, which will continue a tradition of empirical ethics that stretches from Aristotle to Hume to Darwin. 

Another limitation is that in pursuing the simplifications of abstract modelling, behavioralist biopolitics cannot account for the emergent irreducible complexity of political biology.

For example, consider this declaration by Hatemi and McDermott:  "The findings that shared genes can explain up to 50 percent of the variance in political attitudes was a stunning revelation to the discipline (see Alford, Funk, and Hibbing 2005; for an earlier version see Eaves et al. 1999)." 

They don't indicate to the reader, however, that this "stunning revelation" has been challenged by critics who point out that it depends on fundamental assumptions of twin-studies in behavior genetics that contradict recent research showing the variability between twins.  The assumption that monozygotic twins are 100% genetically similar is false.  Because of transposable elements in the human genome ("jumping genes"), copy number variations, variations in mitochondrial DNA, and various epigenetic processes, identical twins are not really identical.  (See the paper by Evan Charney forthcoming in Behavioral and Brain Sciences, which I mentioned in my previous posts on Charney's critique of behavioral biopolitics.)  This biological research suggests that the human genome is so remarkably complex, contingent, and variable, that any genetics research that assumes the fixity of an individual's genome from conception to death must be false.  Hatemi and McDermott say nothing about this research that would cast doubt on their "stunning revelation."

As another illustration of their failure to account for the complexity of human biology, consider the following comment by Hatemi and McDermott:

Biological or genetic processes are not static, nor do they predetermine behavior.  For example, in 2002, Avshalom Caspi and others examined a large sample of male children from  birth to adulthood to determine why only some abused children develop antisocial (violent) behavior, whereas others do not.  They found that a variant of the gene encoding the neurotransmitter metabolizing enzyme Monoamine Oxidase (MAO) moderates the effect of childhood abuse.  Abused children with the genotype that promoted higher levels of MAO activity appeared less likely to become abusers than abused children with the genotype that resulted in lower levels of MAO.  Several additional studies verified their findings, thus partly explaining why not all abused children grow up to victimize others, although others have more recently called their findings into question.  Nonetheless, genotypes can moderate our sensitivity to environmental stimuli along such pathways.

Notice the rhetorical move at the end of this paragraph indicated by that "nonetheless."  Hatemi and McDermott quietly admit that Caspi's findings have been challenged by those who have failed to replicate them.  But then without citing these critics or explaining the debate, Hatemi and McDermott move the reader quickly back to their conclusion favoring Caspi's claims.  What they don't tell the reader is that most of the attempts to replicate Caspi's research have failed, and that this is generally the case with gene association studies, which are notorious for their failure in replication.  This is not a minor point because the acceptance of Caspi's claims is a recurrent theme in this book (see, e.g., 218-19, 256-57).

For some indication of the poor record of genetic association studies, see Marcus Munafo and Jonathan Flint, "Meta-analysis of Genetic Association Studies," Trends in Genetics 20 (September 2004): 439-44; Munafo et al., "Gene x Environment Interactions at the Serotonin Transporter Locus," Biological Psychiatry 65 (2009): 211-19; and Jonathan Flint, Ralph Greenspan, and Kenneth Kendler, How Genes Influence Behavior (Oxford: Oxford University Press, 2010), 76-95.

Another example of how the explanatory and predictive power of behavioral biopolitics is severely limited by the complexity of human biological nature is the research that relies on brain imaging.  Darren Schreiber's chapter on "neuropolitics" in the Hatemi/McDermott book recognizes this problem.

Schreiber identifies a new field of research for biopolitics--"social cognitive and affective neuroscience" (SCAN).  His fundamental assumption is that "all mental activity is at least reflected in the brain" (273).  This is surely uncontroversial for most people.  Most of us would expect that our mental activity is somehow "reflected" in our brains.  The problem, however, is that the connections between mental states and brain states are so complex that it's not clear how we could study this precisely through simplified causal models without distorting the complex reality that we want to explain.

One popular way to study "neuropolitics" is through functional magnetic resonance imaging (fMRI), which appears to give us colorful pictures of the mind at work in the brain, so that we can identify particular parts of the brain with particular mental activities.  So, for example, it seems that we should be able to use fMRI to identify how the brains of voters light up in response to various political candidates, and then political campaign experts can devise techniques for reading the minds of voters through brain imaging studies.  A few years ago, the New York Times published an article entitled "This Is Your Brain on Politics," which claimed, for example, that brain imaging shows that Mitt Romney's image provokes "voter anxiety" as indicated by increased activity in the amygdala.  Claims like this have sparked public excitement about "neuropolitics."

Schreiber explains the techniques of functional MRI research and surveys some of the results.  But while he is optimistic that this can be a powerful tool for political scientists, he rightly warns the reader about the limitations of such research.  The problem is that such brain imaging can be misconstrued as "neophrenology" (274)--conveying the illusion that the colorful pictures from fMRI research actually allow us to read the minds of human beings by identifying mental states with particular brain regions.

It is simply not true, Schreiber admits, that we can clearly identify brain regions as the locations for particular mental activities.  We cannot do this because of the complex interconnectedness of overlapping networks in the brain.  Everything is connected to everything else (290).  Unfortunately, brain imaging studies--especially as they are reported in the popular media--overlook this simple fact of irreducible complexity.

I agree with Schreiber.  But I would go even further in rejecting what I have called "the brain imaging fallacy"--the false assumption that a brain image is a photograph of the mind at work, and that brain imaging explains and predicts thoughts and actions.  Some of my previous posts on this can be found here and here.  Some of the general problems with fMRI are surveyed by Kerri Smith, "fMRI 2.0," Nature 484 (5 April 2012): 24-26.

What Schreiber says about neuroimaging could rightly be said about all of the research surveyed in the Hatemi/McDermott book:

Neuroimaging may provide us evidence that the same region of the brain that enables us to feel pleasure from the spiritual ecstasy of silent prayer (Schjodt et al. 2008) is involved when we punish people who violate social norms (de Quervain et al. 2004) or when we desire a sports car (Erk et al. 2002).  While this data may allow us to make a number of interesting inferences about the pleasurable nature of each of these activities, it does not tell us about the existence of God, the value of a particular social norm, or true quality of the car.  Taking the results of neuroscience seriously does not require us to reject free will (Mele 2009) and may even necessitate the rejection of a merely reductionist project (Mitchell 2009).  The hard questions of political science will remain the hard questions as neuropolitics develops.  However, our field of potential answers will likely be narrowed a bit by the data that this approach generates. (274-75)

Yes, "the hard questions of political science will remain the hard questions" as biopolitics develops.  Even if we cannot find any final answers to those hard questions, we can think through the questions more deeply and weigh the possible answers more rigorously if we integrate behavioral biopolitics into a comprehensive biopolitical science that includes Darwinian psychology, political philosophy, and political history.